SAD - Mini (Situating Prompt): leaderboard
SAD-mini uses portable multiple-choice tasks on model facts, influence and deployment stages. It excludes the self-prediction and self-recognition tasks, so its score does not measure behavioral self-prediction. Uses a situating system prompt reminding the model that it is an LLM.
Metric: Score (%, higher is better). Source: situational-awareness-dataset.org. Saturation forecast: Estimated already saturated. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 Preview (2024-09-12) | 81.28 | #214 |
| 2 | O1 Mini (2024-09-12) | 75.77 | #412 |
| 3 | GPT-4 Preview (0125) | 72.79 | #500 |
| 4 | GPT-4 (0613) | 71.83 | #493 |
| 5 | Claude 3.5 Sonnet | 71.45 | #362 |
| 6 | GPT-4o | 71.35 | #357 |
| 7 | Claude 3 Opus | 70.85 | #565 |
| 8 | Llama 3 Chat 70B | 69.16 | #772 |
| 9 | Claude 3 Sonnet | 65.58 | #774 |
| 10 | Claude 2.1 | 62.69 | #833 |
| 11 | Claude Instant 1.2 | 62.57 | #1003 |
| 12 | Claude 3 Haiku | 61.93 | #801 |
| 13 | Llama 2 70B Chat | 61.87 | #1275 |
| 14 | Llama 2 70B | 58.54 | #1039 |
| 15 | Llama 2 13B Chat | 55.23 | #1414 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=sad-mini-situating-prompt · How It Works · Data refreshed daily, snapshot 2026-11-05.